ContrastMotion: Self-supervised Scene Motion Learning for Large-Scale LiDAR Point Clouds.

Xiangze Jia Hui Zhou Xinge Zhu Yandong Guo Ji Zhang Yuexin Ma

Published in: IJCAI (2023)

Keyphrases

point cloud
image sequences
learning algorithm
three dimensional
laser scanner
stereo camera
high resolution
surface reconstruction
point cloud data
point sets
structure from motion
supervised learning
optical flow
moving objects
motion model
vision system
dynamic scenes
machine learning